Xiangning Chen

I am a researcher at OpenAI working on posttraining and reinforcement learning. I contributed to GPT-5/GPT-5 thinking, o3/o4-mini, GPT-4o posttraining (e.g., reducing sycophancy), next-gen audio models, improved advanced voice mode, etc.

Previously, I co-founded Converge Lab, an AI startup that aims to bring large language models to the physical world.

I obtained my Ph.D. degree at UCLA CS in 2023, advised by Prof. Cho-Jui Hsieh. Prior to UCLA, I received my B.Eng. in 2019 from the Department of Electronic Engineering, Tsinghua University. I've interned at Google Research and Google DeepMind.

Email/ Google Scholar/ Twitter/ LinkedIn

Selected Publications

Red Teaming Language Model Detectors with Language Models

Z. Shi*, Y. Wang*, F. Yin*, X. Chen, K. Chang, C. Hsieh
TACL, 2023
paper / code

Symbol Tuning Improves In-Context Learning in Language Models

J. Wei, L. Hou, A. Lampinen, X. Chen, D. Huang, Y. Tay, X. Chen, Y. Lu, D. Zhou, T. Ma, Q. Le
EMNLP, 2023
paper

Symbolic Discovery of Optimization Algorithms

X. Chen*, C. Liang*, D. Huang, E. Real, K. Wang, Y. Liu, H. Pham, X. Dong, T. Luong, C. Hsieh, Y. Lu, Q. Le
NeurIPS, 2023
paper / code

Lion has been successfully deployed in production systems such as Google’s search ads CTR model
Lion has been widely adopted by the community, e.g., MosaicML employed Lion to train their LLMs

When Vision Transformers Outperform ResNets without Pre-Training or Strong Data Augmentations

X. Chen, C. Hsieh, B. Gong
ICLR (spotlight), 2022
paper

Towards Efficient and Scalable Sharpness-Aware Minimization

Y. Liu, S. Mai, X. Chen, C. Hsieh, Y. You
CVPR, 2022
paper

Rethinking Architecture Selection in Differentiable NAS

R. Wang, M. Cheng, X. Chen, X. Tang, C. Hsieh
ICLR (oral, outstanding paper award), 2021
paper / code

Robust and Accurate Object Detection via Adversarial Learning

X. Chen, C. Xie, M. Tan, L. Zhang, C. Hsieh, B. Gong
CVPR, 2021
paper

DrNAS: Dirichlet Neural Architecture Search

X. Chen*, R. Wang*, M. Cheng*, X. Tang, C. Hsieh
ICLR, 2021
paper / code

Stabilizing Differentiable Architecture Search via Perturbation-Based Regularization

X. Chen, C. Hsieh
ICML, 2020
paper / code

Design and source code adapted from Jon Barron’s site